Chat log from the meeting on 2024-01-30

From OpenSimulator

Jump to: navigation, search
[12:09 PM PT]  Ubit Umarov: welcome :)
[12:09 PM PT]  Ubit Umarov: abotu last week code changes..
[12:10 PM PT]  Ubit Umarov: just a few changes on map, this time also grid side, to save a bit of CO2
[12:10 PM PT]  Ubit Umarov: ie cpu heat.. ie cpu time :)
[12:11 PM PT]  Ubit Umarov: no idea why we where stil using bitmap gfetpixel/setpixel  on grid side
[12:11 PM PT]  Ubit Umarov: well known to be terrible slow
[12:12 PM PT]  Lyr Lobo nods
[12:13 PM PT]  Ubit Umarov: we got a email for joe, about a future ll change on viewer/region messages protocol
[12:13 PM PT]  Ubit Umarov: well..  tpvs need to support what we have..
[12:14 PM PT]  Ubit Umarov: the decision to follow that or not will depend on the respective future, and most likely will only look when it shows up on a viewer for opensim ( yesa potencial chicken-egg issue )
[12:15 PM PT]  Joe Magarac: That's all part of PBR.
[12:15 PM PT]  Ubit Umarov: lludp side a bit more a concern, bc may cause issues with older viewers
[12:15 PM PT]  Joe Magarac: It's live on the SL beta grid now.
[12:15 PM PT]  Jamie.Jordan Hi everybody
[12:15 PM PT]  Ubit Umarov: hmm looked like attachment points, not pbr
[12:15 PM PT]  Ubit Umarov: but dunno
[12:15 PM PT]  Gavin.Hird hi
[12:15 PM PT]  Joe Magarac: The trick is that you don't get those UDP messages unless you ask for a capability that says you understand PBR.
[12:16 PM PT]  Joe Magarac: That's how SL is doing backwards viewer compatibility.
[12:16 PM PT]  Ubit Umarov: well viewers need to adapt to both versions.. even checking packet size on lludp
[12:16 PM PT]  Joe Magarac: Yes, I hit that. Some UDP messages have grown.
[12:16 PM PT]  Ubit Umarov: not currently checking that
[12:17 PM PT]  Ubit Umarov: guess ill need :)
[12:17 PM PT]  Joe Magarac: I do check that in Sharpview, and was logging errrors and wondering where the extra data was coming from.
[12:17 PM PT]  Ubit Umarov: but on pbr there was no need so far
[12:17 PM PT]  Ubit Umarov: well possible not send some overrides may help as transition thing
[12:17 PM PT]  Joe Magarac: This is PBR round 2, which does more on UDP and less on the event poller.
[12:18 PM PT]  Ubit Umarov: fun ll wasn't going to drop UPD ?
[12:18 PM PT]  Joe Magarac: Overrides are coming in as a new UDP generic message.
[12:18 PM PT]  Ubit Umarov: yes getting them here :)
[12:19 PM PT]  Joe Magarac: OK. I don't fully understand PBR at the message level yet.  I'm just passing this info along.
[12:19 PM PT]  Ubit Umarov: well what you shown on the email was about avatar attachment points, whatever those are
[12:19 PM PT]  Ubit Umarov: i mean to need to be sent..
[12:20 PM PT]  Joe Magarac: I'm not sure how avatar attachment points work.
[12:20 PM PT]  Ubit Umarov: well they are fixed things ( as defined on a xml fle )
[12:20 PM PT]  Ubit Umarov: file
[12:21 PM PT]  Ubit Umarov: from the link on your email "More sl-20635 - moved new attachment data to AvatarAppearance message
[12:22 PM PT]  Joe Magarac: Right.
[12:22 PM PT]  Joe Magarac: For now, I'm just passing that along.
[12:22 PM PT]  Ubit Umarov: thanks :)
[12:23 PM PT]  Ubit Umarov: as i said, sadly those changes may mean more compatibily code on viewers
[12:23 PM PT]  Ubit Umarov: not only because we may not support the feature, but also because any change we do takes months to reach grids etc
[12:25 PM PT]  Ubit Umarov: what is their branch  DRTVWR-600-maint-A  ?
[12:25 PM PT]  Joe Magarac: Right. It's better if you don't need a mode switch. Supporting two ways of doing the same thing is better than supporting two viewer versions. If SL started supporting varregions tomorrow, Sharpview would accept them.
[12:26 PM PT]  Ubit Umarov: well whatever.. ll things :)
[12:27 PM PT]  Vincent.Sylvester How is pbr even going to work hypergrid? If I move content to a non-pbr grid it'll strip matovrd from the asset no? Should we then not make hypergrid incompatible to prevent attachments losing pbr data?
[12:27 PM PT]  Joe Magarac: is the link you wanted.
[12:28 PM PT]  Ubit Umarov: oh i see that they did explain what that is to you...
[12:28 PM PT]  Ubit Umarov:
[12:29 PM PT]  Ubit Umarov: in fact on crossings and tps, we wil have issues sending that info, bc attach objects arrive a lot later..
[12:29 PM PT]  Joe Magarac: Right,that's my post on SL forums.
[12:30 PM PT]  Ubit Umarov: so yeah.. that seems to be mostly "noise" TO US
[12:30 PM PT]  Ubit Umarov: to us
[12:30 PM PT]  Gavin.Hird DRTVWR-600-maint-A is more or less only a new main branch. They restarted on A since they had exhausted the alphabet in with the DRTVWR-599-maint-Z. It has not gotten many updates yet, but SL-20635 is one of them
[12:30 PM PT]  Ubit Umarov: well back to current opensim
[12:31 PM PT]  Ubit Umarov: yeah like all new things, teleports specially HG to regions without pbr will be a issue
[12:31 PM PT]  Vincent.Sylvester I'm not sure it justifies a protocol version change though that is why I am asking
[12:31 PM PT]  Ubit Umarov: guess most time, pbr will be ignored on the receiving end
[12:32 PM PT]  Ubit Umarov: and possible another reason to block any appearence chages on HG trips
[12:32 PM PT]  Ubit Umarov: already a but suicidal..
[12:32 PM PT]  Ubit Umarov: and a bit
[12:33 PM PT]  Vincent.Sylvester On one hand it would prevent users from breaking their clothes, but blocking hg just over that for everyone as a result seems aggressive
[12:33 PM PT]  Ubit Umarov: well ppl should knwo that they need to travel light on HG..
[12:33 PM PT]  Ubit Umarov: bento was a issue also :)
[12:34 PM PT]  Ubit Umarov: soem scripts are a issue also..
[12:34 PM PT]  Ubit Umarov: etc
[12:34 PM PT]  Vincent.Sylvester I regularly see hg users start sorting their inventory while elsewhere, which is a workout for the region to do, I think that ship has sailed
[12:34 PM PT]  Vincent.Sylvester We can assume what can break will break
[12:35 PM PT]  Ubit Umarov: well i did add some block of teleports related to the universal wearable etc
[12:35 PM PT]  Ubit Umarov: mb can extend to pbr prims..   duhhh
[12:36 PM PT]  Vincent.Sylvester Gotta wait one way or the other for LL to finally figure out what they want to ultimately do with all these new shiny things
[12:36 PM PT]  Vincent.Sylvester I still don't see a clear direction from them
[12:37 PM PT]  Joe Magarac: The next big LL move is uploading glTF files somehow.
[12:37 PM PT]  Ubit Umarov: they just push fw...
[12:37 PM PT]  Ubit Umarov: that is a nonsense move
[12:37 PM PT]  Ubit Umarov: gltf is a stupid thing to use on streaming
[12:37 PM PT]  Joe Magarac: They're trying to move towards standards.
[12:37 PM PT]  Joe Magarac: No, no, you don't send glTF to the viewer.
[12:37 PM PT]  Orbert.Tatham Unfortunately they don't understand the standards and wnat to do it their own way
[12:38 PM PT]  Ubit Umarov: has they already found, adding that new udp messages
[12:38 PM PT]  Joe Magarac: It's for uploads, instead of COLLADA.
[12:38 PM PT]  Ubit Umarov: well that is just another importer
[12:38 PM PT]  Ubit Umarov: possilbe better a FBX one instead gltf
[12:38 PM PT]  Joe Magarac: FBX isn't a standard, it's just something Autodesk uses.
[12:38 PM PT]  Joe Magarac: glTF is a standard with documents.
[12:39 PM PT]  Ubit Umarov: and use industry standard texture compressors instead on jk2000
[12:39 PM PT]  Ubit Umarov: nahh everyone uses
[12:39 PM PT]  Joe Magarac: Blender and Maya speak glTF, so SL just has to follow along.
[12:39 PM PT]  Ubit Umarov: well a lot more than the users of collada or gltf
[12:40 PM PT]  Vincent.Sylvester Can't say I have made good experiences with gltf in the past, usually seemed like everyone rolled their own interpretation of the standard and nothing was fully compatible...
[12:40 PM PT]  Ubit Umarov: speak means nothing
[12:40 PM PT]  Vincent.Sylvester collada in regards to SL has had similar issues, it has gotten better though
[12:40 PM PT]  Joe Magarac: Everybody renders the "bistro" test scene a little differently .
[12:40 PM PT]  Vincent.Sylvester Still not full to the standard as of last I heard from Beq
[12:41 PM PT]  Ubit Umarov: well importers i fine
[12:41 PM PT]  Joe Magarac: No, there's lots of stuff in glTF that SL doesn't do yet.
[12:41 PM PT]  Ubit Umarov: but what we saw with pbr was the full use of gltf on assrts
[12:41 PM PT]  Ubit Umarov: that is ..  BAD
[12:41 PM PT]  Kayaker Magic: There's lots of stuff in cololada that SL never did
[12:41 PM PT]  Ubit Umarov: gltf is just 2 much verbose.. 2 much junk
[12:42 PM PT]  Vincent.Sylvester SL mantra right there
[12:42 PM PT]  Ubit Umarov: sure a bit less than nonsense xml..
[12:42 PM PT]  Ubit Umarov: of collada
[12:42 PM PT]  Joe Magarac: Yes, shipping glTF to the viewer as JSON inside notation LLSD was a mess.
[12:42 PM PT]  Ubit Umarov: well tey still do it on the asset
[12:42 PM PT]  Ubit Umarov: and on uploads
[12:43 PM PT]  Ubit Umarov: the asset is gltf
[12:43 PM PT]  Ubit Umarov: well a super stripped down gltf
[12:44 PM PT]  Vincent.Sylvester All this just to create even less frames on the viewer end with all the fancy shaders clogging up the renderer
[12:44 PM PT]  Vincent.Sylvester Slideshows!
[12:44 PM PT]  Ubit Umarov: well nvidia wants to sell rtx4090
[12:45 PM PT]  Vincent.Sylvester LL investing in nvidia stock or something?
[12:45 PM PT]  Joe Magarac: No, by the time you get to rendering, it's all low-level mesh.
[12:45 PM PT]  Ubit Umarov: well epic sure seems to
[12:46 PM PT]  Ubit Umarov: unreal 5 is great for the electric power companies also...  nice bills at end of month :P
[12:46 PM PT]  Ubit Umarov: 600w on gpu,  300w on cpu lol
[12:47 PM PT]  Vincent.Sylvester How's that even going to compute when they also push for a mobile viewer... I don't even wanna imagine that
[12:47 PM PT]  Ubit Umarov: wel this fs 7 is working  - ok
[12:47 PM PT]  Ubit Umarov: just a bit more power usage
[12:48 PM PT]  Gavin.Hird I have to go - phone call
[12:48 PM PT]  Ubit Umarov: cya gavin.Hird
[12:48 PM PT]  Vincent.Sylvester Should keep the fingers warm at least huh
[12:48 PM PT]  Ubit Umarov: and pbr is a old thing .. all games have it.. even more complex one
[12:48 PM PT]  Ubit Umarov: ll pbr is a simpler model
[12:49 PM PT]  Vincent.Sylvester It reminds me of like the first games that had mirrors
[12:49 PM PT]  Vincent.Sylvester playstation 2 era
[12:50 PM PT]  Ubit Umarov: wel ofc i hat that ocean water and sky near sun is a mess on this pbre thing
[12:50 PM PT]  Kayaker Magic: I used to warm my office in the winter with an Nvidia Quatro video card. My new GPU uses less power....
[12:50 PM PT]  Ubit Umarov: hate i mean
[12:51 PM PT]  Ubit Umarov: they claim that the 4080 uses less power no?
[12:51 PM PT]  Ubit Umarov: well half the PCI lanes..  only supports 8, etc
[12:51 PM PT]  Ubit Umarov: hmm or 4060?
[12:52 PM PT]  Ubit Umarov: think it is the 4060 sorry
[12:52 PM PT]  Vincent.Sylvester I don't even have an rtx card yet, things just won't go down in price
[12:52 PM PT]  Vincent.Sylvester Things themselves cost as much as the rest of the hardware combined
[12:53 PM PT]  Ubit Umarov: yeah
[12:53 PM PT]  Vincent.Sylvester AI hasn't solved that problem yet... it's actively causing it
[12:53 PM PT]  Ubit Umarov: 2200 euros a 4090 lol
[12:54 PM PT]  Vincent.Sylvester Meanwhile they are also extending lsl at least so we get some useful functions... well we have had them on ossl for a while now
[12:55 PM PT]  Ubit Umarov: wel ll pbr is not the performance eater...  at least i don't notice that much
[12:55 PM PT]  Ubit Umarov: what usefull funtions?
[12:55 PM PT]  Vincent.Sylvester Hashing things
[12:55 PM PT]  Ubit Umarov: those are useful??
[12:56 PM PT]  Vincent.Sylvester Doing auth stuff on http requests and encrypting when you cannot use https
[12:56 PM PT]  Vincent.Sylvester It's something, a little late to the party, but good to have
[12:56 PM PT]  Ubit Umarov: well lsls should not even have http
[12:56 PM PT]  Ubit Umarov: details ;)
[12:56 PM PT]  Vincent.Sylvester I'm waiting on finalizing the wiki pages before starting implementation, cause never know what bugs they have still to fix on those
[12:57 PM PT]  Ubit Umarov: i didn't look to those
[12:57 PM PT]  Ubit Umarov: ofc those things are not light at all
[12:58 PM PT]  Vincent.Sylvester one way hashes are okay, but doing full crypto matching and all that is
[12:58 PM PT]  Vincent.Sylvester Well kinda, the libs in .net for that are not so bad
[12:59 PM PT]  Vincent.Sylvester Seem direct copies of C so pretty close to machine performance
[12:59 PM PT]  Vincent.Sylvester Ideally you don't use those constantly.... ideally
[1:00 PM PT]  Vincent.Sylvester Guess I'll run some performance tests once I get them implemented and maybe compare to SL if I can find a willing guinea pig
[1:03 PM PT]  Lyr Lobo grins
[1:03 PM PT]  Ubit Umarov: :)
[1:03 PM PT]  Lyr Lobo: Have a great day! Must dash to rl
[1:03 PM PT]  Orbert.Tatham Have fun, Lyr
[1:03 PM PT]  Lyr Lobo: thanks!
[1:04 PM PT]  Joe Magarac: Thanks all.
[1:04 PM PT]  Ubit Umarov: cy alyr
[1:04 PM PT]  Ubit Umarov: cya Lyr
[1:04 PM PT]  Ubit Umarov: any other issue?
[1:05 PM PT]  Ubit Umarov: if not, lets end the meeting..  Hope to see you all next week :)
Personal tools
About This Wiki